Browser Bot

An MCP (Model Context Protocol) server for automating Chrome browser operations. It connects to a locally running Chrome instance using the browser_use library, enabling browser automation through natural language instructions.

Features

• Natural language browser control: Automate with instructions like "Search for Python tutorial on Google"
• MCP protocol support: Directly operable from AI assistants such as Claude
• Screenshot capture: Capture visible area or full page screenshots
• Page source retrieval: Useful for HTML structure analysis
• Multiple LLM support: OpenAI GPT and Google Gemini
• Local Chrome connection: Connects to existing Chrome via debug port (:9222)

MCP Tools

Tool	Description
browser_use_local_chrome	Execute browser operations via natural language
get_page_source	Retrieve HTML source of the active tab
get_visible_screenshot	Screenshot of the visible area
get_full_screenshot	Screenshot of the entire page
super_reload	Cache-ignoring reload (CDP / shortcut / JS fallback)

Repository

https://github.com/ytyng/browser-bot